Re: rare crash - FailedAssertion snapbuild.c Line: 580 - Mailing list pgsql-hackers

From Andres Freund
Subject Re: rare crash - FailedAssertion snapbuild.c Line: 580
Date
Msg-id 20180829191526.s7czhujzvcwomh6s@alap3.anarazel.de
Whole thread Raw
In response to rare crash - FailedAssertion snapbuild.c Line: 580  (Erik Rijkers <er@xs4all.nl>)
Responses Re: rare crash - FailedAssertion snapbuild.c Line: 580
List pgsql-hackers
Hi,

On 2018-08-29 17:43:17 +0200, Erik Rijkers wrote:
> To test postgres 11, I still regularly run series of short sessions of
> pgbench-over-logical-replication (basically the same thing that I used last
> year [1] - now in a perl incarnation).  Most of the time the replication is
> stable and finishes correctly but sometimes (rarely) I get:
> 
> TRAP: FailedAssertion("!(TransactionIdPrecedesOrEquals(safeXid,
> snap->xmin))", File: "snapbuild.c", Line: 580)
> 
> This will probably be difficult to reproduce and to act upon but I wanted to
> report it anyway as in the course of the last few months I have seen it
> several times, on several machines. Always rarely, always postgres 11 (I did
> not try other versions).

Thanks for testing! Could you possibly run the tests with core files
enabled, so we at get a backtrace in case of trouble?  Knowing what the
values here are would be tremendously helpful...

Greetings,

Andres Freund


pgsql-hackers by date:

Previous
From: Robert Haas
Date:
Subject: Re: FailedAssertion on partprune
Next
From: Fabien COELHO
Date:
Subject: Re: Removing useless \. at the end of copy in pgbench